(AlRizzi - 05 Oct 2003) I've attached an overall block diagram for the electronics system here.

Quick notes from the first meeting of working group 3.1 (central CPU, power, etc.)

Power basics

• 2 actuators per leg at approx 5W per actuator.
• Bus voltage probably 12V?
• everything else 5V?
• MCUs need 3V -- (downconveters on remote nodes?)
• (Don working w/ Martin to figure bus voltage and batteries)
  • -- DonCampbell? - 12 Sep 2003 Holding for motor selection details. Looks like a 12 or 15v motor will likely be used, maybe an 18v battery.
  • AlRizzi - 12 Sep 2003 - I pushed Martin a bit on this. He was comfortable with us presuming we will use the RE16 motors from Maxon, and assuming that we can work with something like a 3A cont rating on the amps. Still little convergance on what the bus voltage will be -- no one has looked very closely at what our battery stack will be yet.
• (Al is going to work up a rough power budget for all the elecetronics)

• We like the idea of having a centralized power conditioner (think DC->DC converters for driving processors and sensors). Pay attention to trying to use 5V analog electronics for sensosrs and being mindful of noise and reference considerations in board design.

Central CPU

• We favor x86 style cores... Look at the "X-board" and "core modules" at the bottom of RobotCpuBoards
• The plan will be to have a design where we have a "motherboard" with
  • CPU module
  • external communication interface
  • internal robot communication interfaces (3 I2C? buses?)
  • power conditioning for on robot use
  • battery monitoring hardware
• Other on-body sensing will be via I2C? interface
• Need to include debugging/benchtop interfaces for CPU (probably just ethernet, maybe kbd and monitor?).

-- ShaiRevzen? - 10 Sep 2003 - Notice that none of these boards have an I2C? interface; all have multiple USB. This indicates that in order to use the "multiple I2C? bus" remote node architecture we would need to have the USB-to-I2C interface in the robot.

-- AlRizzi? - 10 Sep 2003 - The intent is to support the I2C? infrastructure through our custum I2C? bridges built into the "carrier board" that the CPU module will mount on. This could be done via USB to I2C? conversion, or just through I2C? "masters" hung off of an ISA/LPC bus.

I2C? interface box

We will need to develop a general purpose I2C? interface "box" (e.g. USB to I2C? ) so that we can test remote nodes in a benchtop situation.

Off-robot communication:

• CF based 802.11 stuff (requires a CF carrier on our motherboard -- how hard is that?)
• 900MHz type radio modem modules -- what interface to CPU?
• Bluetooth modules (BlueTooth) -- how hard is this interface and software

• Metrics are SIZE, bandwidth, power and interface complexity/functionality.

• Options starting to be documented in WirelessLinkOptions

Electronics for remote processing

• Hal is a big fan of the 8051 based parts from cygnal

• we are planning on building one remote node per leg including:
  • an MCU
  • power electronics
  • encoder interfaces
  • voltage/current sensing
  • thermal monitoring
  • maybe strain/torque.
• Don is going to look into amplifiers (options at HipAmpOptions)
• block_diagram_2DOF.eps: Hal's initial block diagram
• Al would like to see a pass through connector for the 4-wire I2C? bus that will let the leg and foot stuff easily connect to the robot.
• -- DonCampbell? - 12 Sep 2003 motor_drive_block_don.ps: A different level of detail for a motor drive design. This one deals more with the amp side of things and could easilly be extended to 2 DOF. No part numbers included yet, this is really still just a concept.

Things we will need to design...

Motherboard (complex, diverse functions, power conditioning, I2C? stuff, off robot communication)

• ShaiRevzen? - 10 Sep 2003 - are we sure we want to design one? I would think that buying off the shelf makes much more sense, given the time/risk constraints.

• AlRizzi? - 10 Sep 2003 - I probably used the wrong name for this. Maybe we should call it our carrier/interface board. This is the board that the CPU module will be mounted on, and will include all the cusom and semi-custom electronics we need to interface to the remainder of the robot.

Hip-Nodes (very small, power elctronics and MCU with basic sensing)

Remote-Node template (outline of basics for how to build a remote node)

• HaldunKomsuoglu (09 Sep 2003) - Chris Gaal and I have designed a small circuitry based on C8051F300? for RHexBus development work and as a template for sensor module prototyping. It consists of the MCU itself and some auxiliary components, as well as, connectors for JTAG, RHexBus, and generic I/O (providing the pins of the MCU to the outside world). The board should arrive by Thrusday (11 Sep 2003) and the first electrical debugging should be done by Friday.

USBtoI2C converter (desk/bench-top interface to I2C? modules)

• HaldunKomsuoglu (09 Sep 2003) - There are versions of Cygnal 8051 MCUs with built-in USB 2.0 peripherals. We can implement simple RHexBus? -to-USB bridges based on this component allowing a fast development cycle since this approach would allow us to utilize the RHexBus libraries that would be employed in the physical robot.

• AlRizzi (05 Oct 2003) The more I think about this the more options I realize exist for this. A few options
  1. Build a PC104 (ISA) card with the 4 I2C? interfaces that will eventualy be on the CPU carrier card. Gets us to reuse much of the design (presuming we end up with an ISA interface), but requires us to use a PC104 stack to talk to anything
  2. Use a Cygnal F320 seris part with integrated USB capabilities. This will work fine, but it can only drive 1 I2C? bus. If we want to drive 3-4 buses we will either need to use lots of these, or include a bunch of F300 series parts to act as interfaces to the I2C? buses.

-- AlRizzi? - 09 Sep 2003